Comparison of Brune-type Stress Drops Estimated from Direct S and Coda Waves

Analysis of seismograms recorded by local networks provides fundamental information on earthquake source spectra and site effects. Previous studies have used different parts of the seismograms, e.g., direct P, S, and coda waves, to study these parameters. Site effects and stress drops derived from source spectra have practical implications for strong ground motion prediction. In this study, our aim is to comprehensively compare the source spectra and site effects derived from direct waves and coda waves separately and to constrain the degree of self-similar scaling of earthquake source properties. We analyze about 1500 earthquakes with local magnitude (ML) from 1.2 to 3.5 in the San Jacinto Fault region of California. Both the spectral decomposition method (e.g., Shearer et al., 2006; Trugman & Shearer, 2017) and the improved Coda-Q method (Wang & Shearer, 2017) are applied to estimate source spectra and site effects. To fit the source spectra, we explore both self-similar and non-self-similar Brune-type models. We also study the time-dependent evolution from direct to coda-wave spectra by analyzing spectra from consecutive time windows following the direct S wave and comparing the results with traditional direct S and coda wave analysis. We estimate uncertainties by applying a bootstrap resampling approach. Our goal is to eventually apply these methods in a comprehensive analysis of earthquake source spectra and site effects across southern California.